metatrader5

This MCP server provides a WebSocket-based MCP interface to interact with MetaTrader 5 terminal data and commands, enabling AI applications to subscribe to market data, request account information, and place or manage trades through a real-time channel. The repository also includes a Python client library for convenient interaction and a REST OpenAPI server (via FastAPI) for broader integrations. Start the MCP server to expose a WebSocket API that AI agents can connect to, and use the provided Python client or REST endpoints to access MT5 data and order execution capabilities. The OpenAPI server can be launched with the standard Uvicorn command to expose HTTP endpoints for discovery and interactions, while the MCP WebSocket server handles streaming and command execution.

Prerequisites:

• Python 3.12 or later
• git (optional, for cloning)
• Access to a running MetaTrader 5 terminal

Installation steps:

1. Install Python dependencies pip install -r requirements.txt

2. Run the MCP server (requires MT5 login credentials and server address) python -m metatrader_mcp.cli serve --login <LOGIN> --password <PASS> --server <SERVER>

3. (Optional) Run the OpenAPI server for REST endpoints uvicorn metatrader_openapi.main:app

4. Connect a client (WebSocket or REST) to the appropriate endpoints as documented in the repo

Note: Replace <LOGIN>, <PASS>, and <SERVER> with your MetaTrader 5 account details and target MT5 server. Ensure the MT5 terminal is running and accessible from the machine hosting this MCP server.

Environment variables and configuration tips:

• Keep MT5 terminal running and accessible to the MCP server process.
• If the MT5 login requires two-factor or IP whitelisting, ensure those constraints are handled by the environment where this MCP server runs.
• The MCP server exposes a WebSocket API for real-time data streams and command execution; use the Python client library or REST endpoints to interact.
• If you need to run multiple MT5 accounts, consider duplicating the configuration with distinct login/server pairs and unique server names.
• Check requirements.txt for additional Python dependencies; ensure network access to MT5 terminal and any remote data sources.
• For debugging, start with a minimal login/server and verify connectivity before integrating with AI workflows.